MTH Unboxing: promiseQ

To redu­ce fal­se alarms signi­fi­cant­ly is the task of the Ber­lin-based start­up pro­mi­seQ. Their solu­ti­on is to use the strenths of Arti­fi­cial Intel­li­gence (AI), but to com­ple­ment them with the skills of humans. The two foun­ders Tol­ga Ermis and Eli­as Kar­del told us how exact­ly this is sup­po­sed to work and how the idea came up. Read more in the March issue of MTH Unboxing.

MTH Acce­le­ra­tor: What did you do befo­re promiseQ?

pro­mi­seQ: Tol­ga stu­di­ed indus­tri­al engi­nee­ring, focu­sing on pro­duct deve­lo­p­ment. Eli­as stu­di­ed electri­cal engi­nee­ring, focu­sing on new tech­no­lo­gies and algo­rith­ms.
We met at HEL­LA Agla­ia, a Volks­wa­gen sub­si­dia­ry that deve­lo­ps com­pu­ter visi­on for cars. Our team was working to inte­gra­te arti­fi­cial intel­li­gence into a ran­ge of sen­sors and inter­faces used in came­ra-based dri­ving func­tions such as bra­ke assist, lane kee­ping, and mul­ti-touch came­ras. Working in the final pha­se of the deve­lo­p­ment pipe­line, we were able to meti­cu­lous­ly exami­ne (and admi­re) all the abi­li­ties of arti­fi­cial intel­li­gence. Howe­ver, we also beca­me awa­re of the nume­rous errors and safe­ty gaps that can ari­se from a lack of moni­to­ring.
As part of a pilot pro­ject, we inves­ti­ga­ted the bene­fits of human invol­ve­ment in cases when the ana­ly­ses by arti­fi­cial intel­li­gence were unsuc­cess­ful (for examp­le, reco­gni­zing an onco­m­ing vehi­cle or a road sign). The pro­ject was even­tual­ly ter­mi­na­ted by the lea­ders at HEL­LA, but we had dis­co­ve­r­ed some­thing gre­at during our time working on it.
Sub­se­quent­ly, Eli­as left for Tai­wan, and Tol­ga took a job at IAV, ano­t­her Volks­wa­gen sub­si­dia­ry. The­re, Tol­ga con­duc­ted test dri­ves to eva­lua­te front came­ra sen­sing. The bene­fits of human invol­ve­ment in such pro­ces­ses were unmistaka­ble, so we star­ted dis­cus­sing the issue again. We tal­ked about how to app­ly crowd­wor­king to pro­blems in are­as such as qua­li­ty con­trol and healthcare.

MTH Acce­le­ra­tor: What do you do in your free time?

Tol­ga: I do a lot of sports to reb­a­lan­ce mys­elf (HIIT, jog­ging, Thai boxing, BJJ––I have a black belt in Tae­kwon­do). I also like to read and watch movies when I have the time.
Eli­as: I love Asia. I lived in Asia for a year and a half, and I prac­ti­ce Chi­ne­se sword figh­t­ing. After a long day of soft­ware deve­lo­p­ment, I also like to work with my hands in my (hob­by!) wood workshop

MTH Acce­le­ra­tor: What is promiseQ?

pro­mi­seQ: pro­mi­seQ is a tech start­up foun­ded in Ber­lin. We have deve­lo­ped a cloud-based plat­form that veri­fies and fil­ters fal­se alarms in real time through AI and an on-demand human-in-loop ser­vice. The pro­mi­seQ hybrid engi­ne com­ple­ments AI, which con­ti­nuous­ly learns with human-in-loop veri­fi­ca­ti­on. This method allows us to take advan­ta­ge of both worlds: machi­nes and humans.
The incre­a­sing popu­la­ri­ty of crowd­work and modern cloud tech­no­lo­gies allows us to sca­le as nee­ded. In addi­ti­on, by using the human-in-loop approach, we can gain a signi­fi­cant com­pe­ti­ti­ve edge for various sce­n­a­ri­os out­side the scope of tra­di­tio­nal secu­ri­ty app­li­ca­ti­ons. The pro­mi­seQ hybrid engi­ne can be used to moni­tor busi­ness pro­to­cols and pro­ces­ses (i.e. cor­rect beha­vi­or in the cor­rect order), detect hazar­dous situa­tions (i.e. acci­dents, fights, theft), moni­tor inven­to­ry, detect con­ta­mi­na­ti­on in the food and cos­me­tics indus­try, detect wild­life on farms, pro­tect arti­facts in muse­ums, and much more. The pos­si­bi­li­ties of promiseQ’s hybrid approach are endless!

MTH Acce­le­ra­tor: Why the MTH Accelerator?

pro­mi­seQ: Through men­to­ring and coa­ching, MTH Acce­le­ra­tor has given us the oppor­tu­ni­ty to impro­ve our acti­vi­ties in busi­ness deve­lo­p­ment and sales. We have also been able to expand our net­work with MTH Accelerator’s com­mu­ni­ty of sup­por­ters and alum­ni. This in turn sup­ports our fund­rai­sing activities.

MTH Acce­le­ra­tor: Why promiseQ?

pro­mi­seQ: Right now, crowd­wor­king as a way to pro­vi­de secu­ri­ty is still a novel­ty in our rather con­ser­va­ti­ve indus­try. In a world of sur­veil­lan­ce cen­ters and auto­ma­ted solu­ti­ons, crowd­wor­king to pro­tect an infra­st­ruc­tu­re is a com­ple­te­ly new idea. But the num­bers speak for them­sel­ves: in cer­ti­fied pilot pro­jects, pro­mi­seQ has achie­ved a 99.5% suc­cess rate in redu­cing fal­se alarms and has done so at a low cost. The sug­ges­ti­on of decen­tra­li­zed secu­ri­ty moni­to­ring may sound far-fetched––just like the idea did at one time that cars could move fas­ter than horses––but the suc­cess rates and cost savings show us that it is anything but.

MTH Acce­le­ra­tor: What is your mission?

pro­mi­seQ: Our mis­si­on is to redu­ce or eli­mi­na­te the num­ber of inco­m­ing fal­se alarms so that our cli­ents can focus on their core busi­ness of crea­ting secu­ri­ty and adding value for their customers.

About the MTH Unboxing

The MTH Acce­le­ra­tor port­fo­lio is gro­wing and gro­wing. But what exact­ly do the teams do and, abo­ve all, what did they do befo­re they beca­me entre­pre­neurs? In MTH Unboxing, our new inter­view seri­es, we shed some light on the sub­ject and intro­du­ce the teams in more detail. MTH Acce­le­ra­tor Babels­berg is the pro­gram for star­tups and foun­ding teams from the metro­po­li­tan regi­on who are loo­king to explo­re tech­no­lo­gies and digi­tal busi­ness models and take their com­pa­nies to the next level.

Learn more about the pro­gram, our latest port­fo­lio star­tups and events in our newsletter.